Abstract:

Prediction of spatial distribution and occurrence area of alien forest pests and diseases is a prerequisite for making managing measures of biological invasions. In this paper data from 75 pine wilt disease occurrence points with geographic coordinates and 68 environmental variables were gathered in 2007. Four habitat modeling methods of Classification and Regression Trees(CART),Genetic Algorithm for Rule-set prediction(GARP),maximum entropy method (Maxent),and Logistic Regression(LR) were introduced to generate potential geographic distribution maps for invasions of pine wood nematode in Jiangsu province. Then the occurrence area was predicted in each county of Jiangsu province. Results showed that CART outperformed other three models. Slope, precipitation seasonality (bio15), compound topographic index (CTI), mean temperature of driest quarter (bio9), slope aspect, maximum temperature of warmest month (bio5) were the six forcing environmental factors. CART model predicted that future occurrence area of pine wilt disease would account for 39.04% of total pine forest, 2.73 times of present infected area of the pest. Yixing, Liyang and Jurong would be the most susceptible counties to the pest, while Yixing,Liyang and Nanjing urban would have the largest infected forest area.

Key words: Bursaphelenchus xylophilus, potential habitat, prediction, Jiangsu Province